I want to explore the relationship between frequency of social media use (predictor) with high or low levels of self-esteem (criterion 1) and high or low levels of social comparisons (criterion 2). I'm doing a correlation analysis but was also thinking of doing regression as well. What type of regression would I need to use? Is it necessary? I'm confused.
Hi
I do not know what question you are going to answer that you chose this analysis method, if it were so we could better discuss the type of analysis needed. But in my opinion, canonical correlation is one of the appropriate ways to understand the role of predictor and criterion variables.
It depends on the nature of the the two DVs. If you describe them better it will be easier to help you. Best wishes David Booth
If the only variables in your regression are the two that you used in each correlation, then the results will identical. The main advantage of the regression would be to introduce control variables, so that you can assess the effects of your independent variable, above and beyond the effects of those control variables.
Sounds like you probably have a mixture of random and fixed effects. I might suggest having a look at linear mixed effects models. These models can fit both continuous (media use) and categorical (self esteem, social comparisons) variables. lme4 is a nice R package to use and there are several nice free tutorials on the web.
- Badges
- Science topic
- Similar topics
- Counseling Psychology
- Confusion